associate editor: d. shugar strategies toward the design ... · model led to the identification and...

Pharmacol. Ther. Vol. 82, Nos. 2–3, pp. 195–206, 1999Copyright © 1999 Elsevier Science Inc.

ISSN 0163-7258/99 $–see front matterPII S0163-7258(98)00044-8

Associate Editor: D. Shugar

Strategies toward the Design of Novel and Selective Protein Tyrosine Kinase Inhibitors

Peter Traxler* and Pascal Furet

NOVARTIS PHARMACEUTICALS, THERAPEUTIC AREA ONCOLOGY, NOVARTIS LIMITED, CH-4002 BASEL, SWITZERLAND

ABSTRACT. Protein tyrosine kinases play a fundamental role in signal transduction pathways. Deregulatedtyrosine kinase activity has been observed in many proliferative diseases (e.g., cancer, psoriasis, restenosis,etc.). Tyrosine kinases are, therefore, attractive targets for the design of new therapeutic agents againstcancer. We have built up a pharmacophore model of the ATP-binding site of the epidermal growth factorreceptor (EGFR) kinase and used it for the rational design of kinase inhibitors. Several examples of thesuccessful use of this model are presented in this review. Amongst these, 4-substituted-pyrrolo[2,3-

d

]pyrimidines, a new class of highly potent and selective inhibitors of the EGFR kinase, have been identifiedand further optimized. The most active derivatives inhibited the EGFR tyrosine kinase with IC

50

valuesbetween 1 and 5 nM. In EGF-dependent cellular systems, tyrosine phosphorylation, as well as c-

fos

mRNAexpression, was inhibited with similar IC

50

values. Further successful application of this pharmacophoremodel led to the identification and optimization of phenylamino-pyrazolo[4,3-

d

]pyrimidines and substitutedisoflavones and quinolones, other classes of potent, selective, and ATP competitive EGFR kinase inhibitorswith IC

50

values in the low nanomolar range. Structure-activity relationships of both classes are discussed.

pharmacol. ther.

82(2–3):195–206, 1999.

© 1999 Elsevier Science Inc. All rights reserved.

KEY WORDS.

Epidermal growth factor receptor tyrosine kinase, platelet-derived growth factor receptortyrosine kinase, pharmacophore model, ATP competitive inhibitors, protein tyrosine kinase inhibitors.

CONTENTS1. I

NTRODUCTION

. . . . . . . . . . . . . 1952. T

HE

ATP-B

INDING

S

ITE

OF

T

YROSINE

K

INASES

AS

A

T

ARGET

FOR

D

RUG

D

ISCOVERY

. . . . . . . . . . . . 1963. P

HARMACOPHORE

M

ODEL

OF

THE

ATP-B

INDING

S

ITE

OF

P

ROTEIN

K

INASES

. . 1974. 4-S

UBSTITUTED

P

YRROLO

[2,3-

D

]

PYRIMIDINES

. . . . . . 199

5. 4-P

HENYLAMINO

-P

YRAZOLO

[3,4-

D

]

PYRIMIDINES

. . . . . 2006. I

SOFLAVONES

AND

3-P

HENYL

-4(1

H

)-Q

UINOLONES

. . . . . 2027. P

HENYLAMINO

-

PYRIMIDINES

. . . . . . . 2048. O

UTLOOK

. . . . . . . . . . . . . . . . . 205R

EFERENCES

. . . . . . . . . . . . . . . . . 205

ABBREVIATIONS.

cAMP, cyclic AMP; CDK, cyclin-dependent kinase; EGFR, epidermal growth factorreceptor; FGFR, fibroblast growth factor receptor; IGF-IR, insulin-like growth factor-I receptor; IR, insulinreceptor; PDGFR, platelet-derived growth factor receptor; PKC, protein kinase C; PTK, protein tyrosine kinase;SAR, structure-activity relationship; VEGFR, vascular endothelial growth factor receptor.

1. INTRODUCTION

Protein kinases play an important role in signal transduc-tion pathways regulating a number of cellular functions,such as cell growth, differentiation, and cell death. A vari-ety of tumor types have dysfunctional growth factor recep-tor tyrosine kinases, resulting in inappropriate mitogenicsignaling. Protein tyrosine kinases (PTKs) are, therefore,attractive targets in the search for therapeutic agents, notonly against cancer, but also against many other diseases.

About 10 years ago, when research in the signal trans-duction field was initiated in many pharmaceutical compa-nies, the EGFR tyrosine kinase was one of the first tyrosinekinases described in the literature, and, therefore, was cho-sen as a target to start drug discovery projects. Meanwhile,many other receptor and nonreceptor tyrosine kinases havebeen described and found to be attractive targets for re-

search programs in cancer and noncancer indications.Novel members of the family of the EGFR PTK, such as the

p185

c-

erbB2

PTK(c-

erbB

2 gene product) and the c-

erb

B3 andc-

erb

B4 gene products, have been identified and are used astargets for medicinal chemistry programs. In the last fewyears, a switch from the EGFR target to other receptor andnonreceptor tyrosine kinase targets, such as the platelet-derived growth factor receptor (PDGFR), vascular EGFR(VEGFR), fibroblast growth factor receptor (FGFR), insu-lin-like growth factor-I receptor (IGF-IR), insulin receptor(IR), Abl, and Src family tyrosine kinases, has been ob-served. Additional targets under evaluation also include thec-Met, Trk, or the cytoplasmic Jak2 kinases.

In the last decade, hundreds, or even thousands, of ty-rosine kinase inhibitors have been described and reviewedin the literature (Bridges, 1995; Burke, 1992; Fry, 1994;Levitzky and Gazit, 1995; Spada and Myers, 1995; Traxlerand Lydon, 1995; Traxler, 1997; Traxler, 1998). Most ofthem only served as tools to set up

in vitro

assay systems and

*Corresponding author’s address: Novartis Pharmaceuticals, TherapeuticArea Oncology, K136.4.94, Novartis Limited, CH-4002 Basel, Switzerland.

196

P. Traxler and P. Furet

to prove their functionality; only a few inhibitors demon-strated

in vivo

efficacy in relevant tumor models in nudemice. The first compounds, in most cases, inhibitors of theEGFR tyrosine kinases, to enter preclinical developmentfailed to fulfill the pharmacodynamic (efficacy, selectivity),pharmacokinetic (resorption, metabolism), toxicological,and technical (synthesis, formulation) requirements, andnever entered clinical trials. Meanwhile, some of these hur-dles have been overcome. Currently, there are seven low-molecular weight compounds described as more or less se-lective tyrosine kinase inhibitors in early phases of clinicaltrials (Fig. 1). In addition, several candidates are in late-

stage preclinical development and are close to enteringPhase I trials.

2. THE ATP-BINDING SITE OF TYROSINE KINASES AS A TARGET FOR DRUG DISCOVERY

Within the great number of different structural classes oftyrosine kinase inhibitors, compounds competing withATP for binding at the catalytic domain of the kinase areconsidered to be of special interest. Despite the fact thatcatalytic domains of most protein kinases share significantamino acid sequence homology and conserved core struc-

FIGURE 1. Tyrosine kinase inhibitors in clinical trials. Compounds marked with an asterisk are ATP competitive inhibitors.

Design of Novel and Selective Tyrosine Kinase Inhibitors

197

tures, it is now accepted that the ATP-binding site of pro-tein kinases is an exciting target for rational drug design. Infact, out of the seven tyrosine kinase inhibitors that are cur-rently in clinical trials, five are described to be ATP com-petitive inhibitors (Fig. 1). Numerous compounds of struc-turally diverse classes have proven to be highly potent andselective ATP competitive tyrosine kinase inhibitors. Spe-cial interest has focused on a special group of compoundscontaining a phenylamino-pyrimidine moiety in theirstructure, such as phenylamino-quinazolines (Fry

et al.

,1994; Rewcastle

et al.

, 1995; Ward

et al.

, 1994), phenyl-amino-pyrido[

d

]pyrimidines (Rewcastle

et al.

, 1996; Thomp-son

et al.

, 1995), phenylamino-pyrimido[5,4-

d

]pyrimidines(Rewcastle

et al.

, 1997), phenylamino-pyrrolo[2,3-

d

]pyrim-idines (Traxler

et al.

, 1996), and phenylamino-pyra-zolo[3,4-

d

]pyrimidines (Traxler

et al.

, 1997).

3. PHARMACOPHORE MODEL OF THEATP-BINDING SITE OF PROTEIN KINASES

Although more than 30 crystal structures of protein kinasescomplexed with ATP or an ATP competitive inhibitor

have been published in the last few years, many attempts toobtain a crystal structure of the EGFR PTK have not beensuccessful. Furet

et al.

(1995) published the first data of apharmacophore model for inhibitors competing for theATP-binding site of the EGFR PTK. In the course of thiswork, a hypothetical model for the binding mode of theprotein kinase inhibitor staurosporine in the ATP-bindingsite of the cyclic AMP (cAMP)-dependent protein kinasewas developed by using the published crystal data of this ki-nase (Zheng

et al.

, 1993). Recently, a published crystalstructure of a complex of staurosporine with cyclin-depen-dent kinase (CDK)2 fully confirmed the proposed bindingmode of staurosporine (Lawie

et al.

, 1997). In addition, ourbinding hypothesis was fully consistent with a model of theEGFR-PTK constructed by homology to the X-ray crystalstructure of the cAMP-dependent protein kinase. Thispharmacophore model has then been used successfully forthe design and synthesis of 4-phenylamino-pyrrolo[2,3-

d


et al.

, 1996) and 4-phenylamino-pyrazolo[3,4-

d


et al.

, 1997). As an ex-ample, the binding mode of 4-phenylamino-pyrrolo[2,3-

d

]pyrimidine CGP 59326 (Compound

1

) is shown in Fig. 2.

FIGURE 2. CGP 59326 docked at the ATP-binding site of the EGFR kinase. Sulfur-aromatic interaction between the chloro-phenylring and Cys773 of the sugar pocket.

198

P. Traxler and P. Furet

Based on our own experience, together with the availabilityof published crystal data of several protein kinases, themodel was refined and is now of general usefulness for therational design of protein kinase inhibitors (Fig. 3). Ac-cording to this model, the ATP-binding site in protein ki-nases can be divided into five regions (P. Furet, manuscriptin preparation):

Adenine region:

This region is of mostly hydrophobic char-acter. In the EGFR kinase, the N1 and N6 nitrogens ofthe adenine ring of ATP are engaged in a hydrogen bonddonor-acceptor system with two amino acid residues,Gln 767 and Met 769, of the hinge region, correspond-ing to Glu 121 (backbone carbonyl) and Val 123 (back-bone NH) in cAMP-dependent protein kinase and an-chor the nucleotide. In the pyrrolo-pyrimidine CGP59326 (Fig. 2), this hydrogen bond donor-acceptor sys-tem is formed by the pyrrolo NH (H-donor) and the N1of the pyrimidine ring (H-acceptor) (Traxler

et al.

,1996). Many other potent inhibitors use at least one ofthese two hydrogen bonds (e.g., quinazolines). Althoughnot used by ATP, the backbone carbonyl of the residuecorresponding to Val 123 in cAMP-dependent proteinkinase can also serve as a hydrogen bond acceptor for in-hibitor binding, as shown by X-ray crystallography for

the CDK2 inhibitor olomoucine (Schulze-Gahmen

etal.

, 1995).

Sugar pocket:

This pocket is of hydrophilic character inmost protein kinases. It has been exploited in the designof potent and selective EGFR kinase inhibitors of thepyrrolo-pyrimidine or pyrazolo-pyrimidine classes. In CGP59326 (Fig. 2), a chlorophenyl moiety replaces ribose ofATP. In addition, a sulfur-aromatic interaction betweenthis moiety and residue Cys 773 of the pocket is assumed(Traxler

et al.

, 1996). This Cys 773 is unique to theEGFR family of kinases and provides both potency andselectivity.

Hydrophobic region I:

This pocket, extending in the di-rection of the lone pair of the N7 nitrogen of adenine, ispresent, but not used, by ATP in most protein kinases,thereby offering the medicinal chemist many possibili-ties for the design of new inhibitors. In the EGFR kinase,its size is controlled by two main amino acid residues,Thr 766 and Thr 860, thus playing an important role ininhibitor selectivity. In other protein kinases, thispocket is relatively small (e.g., in CDKs, a bulky pheny-lalanine [Phe 80 in CDK2] occupies the position corre-sponding to Thr 766 in the EGFR kinase). Many potentserine/threonine or tyrosine kinase inhibitors occupy thispocket (e.g., pyrrolo-pyrimidines, pyrazolo-pyrimidines,

FIGURE 3. Protein kinase pharmacophore.

Design of Novel and Selective Tyrosine Kinase Inhibitors

199

phenylamino-quinazolines, pyrido-pyrimidines, pyrimido-pyrimidines, oxindoles, etc.). In CGP 59326, the twomethyl groups of the pyrrole ring are pointing towardsthis pocket, thereby providing only a suboptimal occu-pancy of the available space.

Hydrophobic region II:

This region corresponds more to ahydrophobic slot opened to solvent, which is not used byATP and can be exploited to gain binding affinity. Inthe EGFR kinase, it is formed by residues Leu 694 andGly 772. The X-ray crystal structures of indolinone in-hibitors in complex with the FGFR kinase exploitingthis slot have been published recently (Mohammadi

etal.

, 1997). In certain other protein kinases (e.g., mito-gen-activated protein and CDK family kinases), the shapeof the slot is slightly altered due to the deletion of theamino acid corresponding to Gly 772 in the EGFR kinase.

Phosphate binding region:

This region has high solventexposure and is not of primary importance with respectto binding affinity. However, it can be useful to improvethe selectivity of inhibitors since it contains noncon-served amino acids.

Independently, Parke-Davis (Ann Arbor, MI, USA) sci-entists proposed a model for the binding of 4-phenylamino-quinazolines (Palmer

et al.

, 1997) and pyrido[2,3-

d

]pyrim-idines (Trumpp-Kallmeyer

et al.

, 1998) at the ATP-bindingsite of the EGFR kinase. While there is no other consistentway than their proposed binding mode to accommodate thepyrido[2,3-

d

]pyrimidine inhibitors in the ATP-binding siteof a tyrosine kinase, this is not the case for the phenyl-amino-quinazolines. According to their model, the phenyl-amino moiety of these inhibitors occupies hydrophobic re-gion I of the ATP-binding site, while the N1 nitrogen of

the pyrimidine ring forms a hydrogen bond with the back-bone NH of the residue (Met769) corresponding to Val123in cAMP-dependent protein kinase. Assuming the samehydrogen bond interaction, many of these inhibitors (withthe exception of pyrazolo-quinazoline analogues alkylatedat N1) can also be accommodated in a binding mode wherethe phenylamino moiety occupies the sugar pocket, as wehypothesize for our pyrrolo-pyrimidine inhibitors. The con-cept that kinase inhibitors belonging to a given chemicalclass adopt a unique binding mode is not supported by re-cent X-ray crystallographic structures of purine inhibitorsin complex with CDK2 (Schulze-Gahmen

et al.

, 1995). De-pending on the types and positions of substituents on agiven kinase inhibitor template, changes in the bindingmode may occur.

In the following sections, examples of the successful ap-plication of our pharmacophore model are described.

4. 4-SUBSTITUTED-PYRROLO[2,3-

d

]PYRIMIDINES

The first successful application of the pharmacophoremodel for ATP competitive inhibitors interacting with theactive site of the EGFR tyrosine kinase led to the identifi-cation of 4-phenylamino-7

H-

pyrrolo[2,3-

d

]pyrimidines as anovel class of EGFR PTK inhibitors. In an interactive pro-cess, this class of compounds was then further optimized(Traxler

et al.

, 1996). In the structure-activity relationship(SAR) study of a first series of compounds (Compound

1

,Table 1), the influence of substituents in the anilino moietywas evaluated. The same preference for electron-withdraw-ing groups (e.g., Cl or Br) at the 3-position was shown, asalready described for 4-phenylamino-quinazolines (Rewcas-tle

et al.

, 1995) or 4-phenylamino-pyrido-pyrimidine series(Rewcastle

et al.

, 1996).In a further series, the large hydrophobic pocket, as sug-

gested by the pharmacophore model, was explored by re-placement of either one or both methyl groups in positions5 and 6 of the pyrrole ring by bulkier substituents (Com-pounds

2

–

7

, Table 1). In general, bulky lipophilic groupswere tolerated in both positions. Replacement of the 5-methylgroup or of both methyl groups by a phenyl moiety (Com-pounds

3

and

4

) only led to a slight decrease of activity,whereas the 6-phenyl analogue

2

was slightly more activecompared with Compound

1

. However, a limitation wasreached, with Compound

5

bearing a biphenyl moiety inposition 6 of the pyrrole ring. This compound was inactive.Replacement of both methyl groups by a cyclohexyl ring re-tained potency (Compound

6

). Oxidation of the cyclo-hexyl ring in

6

to the indolo-pyrimidine derivative

7 fur-ther increased the activity against the EGFR kinase. Withan IC50 of 6 nM, this compound was equipotent to the3-bromophenylamino-quinazoline PD 153035 (Fry et al.,1994; Rewcastle et al., 1995), which, under our assay condi-tions, had an IC50 value of 6 nM. Although more potentthan Compound 1, Compound 7 was not evaluated furtherdue to its unsatisfactory solubility profile. Instead, 1 (CGP

TABLE 1. SAR of Pyrrolo-Pyrimidine Derivatives

Compound R1 R2

EGFR(IC50 mM)

CGP 593261 (CGP 59326) CH3 CH3 0.0272 C6H5 CH3 0.0153 CH3 C6H5 0.2304 C6H5 C6H5 0.0965 Biphenyl CH3 .1006 0.0297 0.006

PD 153035 0.006

200 P. Traxler and P. Furet

59326) was selected as a development candidate. CGP59326 inhibited the EGFR tyrosine kinase with an IC50

value of 27 nM. Kinetic analysis revealed competitive typekinetics relative to ATP. High selectivity towards a panelof receptor and nonreceptor tyrosine kinases, as well asserine/threonine kinases (protein kinase C [PKCa], PKA),was observed. In cells, EGF-stimulated cellular tyrosinephosphorylation was inhibited by CGP 59326 at an IC50

concentration of 0.3 mM, whereas the ligand-induced re-ceptor autophosphorylation of the PDGFR was not affectedby concentrations of up to 100 mM. Furthermore, 1 wasable to selectively inhibit c-fos mRNA expression in EGF-dependent cell lines (IC50 ,0.1–1 mM), but not in EGF-independent cell systems (IC50 . 100 mM) (Traxler et al.,1996). When tested on a number of EGF-overexpressingepithelial cell lines (NCI-H596, MDA-MB468, A431,BALB/MK, SK-BR3, BT20, etc.), CGP 59326 inhibitedtheir proliferation with IC50 values between 0.5 and 1.9mM, while having only weak antiproliferative activityagainst EGFR negative cell lines (e.g., NCI-H520, NCI-H69, FDC-P1, etc.).1 A moderate antitumor effect was de-scribed with 1 in an orthotopic bladder carcinoma (253JB-V)in nude mice at oral doses between 10 and 40 mg/kg. In ad-dition, inhibition of activated EGFR, down-regulation ofVEGF in the plasma, and inhibition of metastasis was alsoseen in these experiments.2 In pharmacokinetic studies,good bioavailability with high plasma levels was demon-strated after oral administration of the compound to mice,rats, and dogs (P. Traxler et al., unpublished results). CGP59326 is an interesting candidate for further evaluations.

Further optimization of the pyrrolo-pyrimidine structureby attachment of numerous substituents in the 6-position ofthe pyrrole ring led to a series of new derivatives with im-proved biological and physico-chemical properties (Table 2).This includes compounds with acids, ester, or amide groups(e.g., Compound 8); heterocyclic rings (e.g., Compound9); and especially meta- or para-substituted aromatic rings(Compounds 10–14). Many of these compounds were morepotent than CGP 59326 and had IC50 values between 1 and5 nM against the EGFR kinase. Finally, further optimiza-tion at the meta- or para-substituent of the aromatic ringand replacement of the m-chloro-anilino moiety at the 4-position of the pyrimidine ring by an (R)-phenethylaminomoiety, further improved potency and pharmacokinetic be-haviour of the compounds (Compounds 15-22) (Table 3).The most interesting compounds of this series inhibited the

EGFR kinase with IC50 values between 1 and 5 nM and hada selectivity ratio of .1000 against most tested tyrosine andthreonine/serine kinases. Interestingly, the correspondingenantiomeric analogue of 16 with an (S)-phenethylaminomoiety was only weakly active (IC50 .0.2 mM). In EGFR-overexpressing A 431 cells, EGF-stimulated phosphoryla-tion was blocked with IC50 values between 10 and 50 nM(Compounds 15–18, 20, and 21), whereas PDGF-inducedphosphorylation was not inhibited (IC50 .10 mM), thus in-dicating high selectivity for the inhibition of the ligand-activated EGFR signal transduction pathway. When testedfor antiproliferative activity, these compounds inhibitedproliferation of the EGFR-dependent BALB/MK cell linewith IC50 values between 0.1 and 0.4 mM, but were onlyweakly active (IC50 .10 mM) against a panel of EGF-inde-pendent cell lines (e.g., FDC-P1 or T24) (Traxler et al.,1998). Extended biological profiling of selected developmentcandidates is currently ongoing at Novartis.

5. 4-PHENYLAMINO-PYRAZOLO[3,4-d]PYRIMIDINES

Encouraged by the successful application of our pharma-cophore model to the class of the pyrrolo-pyrimidines, weapplied it on the pyrazolo-pyrimidines 23 and 24, which


Compound R1 R2

EGF-R(IC50 mM)

1 CH3 CH3 0.0278 CONHCH3 H 0.003

9 H 0.007

10 H 0.015

11 H 0.003

12 H 0.003

13 H 0.004

14 H 0.001

1Mett, H., Buchdunger, E., Cozens, R., Stover, D. and Lydon, N. (1998)CGP 59326, a potent protein tyrosine kinase inhibitor which selectivelyblocks growth of epidermal growth factor receptor expressing tumor cells.In: 89th Annual Meeting of the American Association for CancerResearch, New Orleans, LA, March 28–April 1, Poster Discussion, 3809.

2Perrotte, P., Kim, Y. K., Knighton, B., Radinsky, R., Killion, J. J. andDinney, C. P. N. (1998) Oral administration of a new epidermal growthfactor receptor protein tyrosine kinase inhibitor, CGP 59326, inhibitsgrowth of human transitional cell carcinoma (TCC) of the bladder. In:89th Annual Meeting of the American Association for Cancer Research,New Orleans, LA, March 28–April 1, Poster Discussion, 2153.

Design of Novel and Selective Tyrosine Kinase Inhibitors 201

were identified in a random screening program as potent in-hibitors of the EGFR tyrosine kinase, with IC50 values of0.22 and 1.2 mM, respectively. In fact, we were able to findan application of our model for both structures (dual fit,Fig. 4). Superimposition of structure 24 with ATP wouldsuggest the bidentate-hydrogen-bond-donor-acceptor sys-tem for the 4-amino group (donor) and the N(5) pyrimi-dine nitrogen (acceptor). In this binding mode, the phenylring attached to the N1 of the pyrazole ring would replace

ribose of ATP in the “sugar pocket,” whereas the phenylsubstituent at the C3 position of the pyrazole ring points to-wards the large hydrophobic pocket (Fig. 4) (Traxler et al.,1997). Recently, Pfizer (Groton, CT, USA) scientists de-scribed a series of 4-amino-pyrazolo[3,4-d]pyrimidines cor-responding to this binding mode, with substituents at thepyrazole nitrogen and an aromatic ring at the 3-position ofthe pyrazole ring. An analogue thereof with a tertiary butylgroup at the pyrazole nitrogen was described to be a potentinhibitor of the Src family of tyrosine kinases (Lck and Fyn)(Hanke et al., 1996).

Assuming a similar binding mode of pyrazole 23 (or itsequipotent 6-desamino derivative), as has already beendemonstrated with 4-phenylamino-pyrrolo-pyrimidines (Trax-ler et al., 1996), would imply that in this case, (1) the NH1of the pyrazole ring and N7 of the pyrimidine ring form thebidentate-hydrogen-bond-donor-acceptor system, (2) the4-amino group points toward the “sugar pocket” not yet fill-ing it, and (3) the anilino substituent at the C3 position ofthe pyrazole ring again is filling the large hydrophobicpocket. This binding mode suggests the addition of a phe-nyl or meta-chlorophenyl moiety, respectively, to the 4-amino group of the pyrimidine ring of the lead Compound23, thus leading to the pyrazolo-pyrimidine analogue 25 asa first target structure (Fig. 4). In fact, Compound 25 hadan IC50 value of 0.033 mM and was almost one order ofmagnitude more potent than the parent Compound 23(Traxler et al., 1997). Optimization of this structural classconcentrated mainly on derivatives with various substitu-ents at the C3 position of the pyrazole ring, thereby exploit-ing the cavity of the large hydrophobic pocket of the targetenzyme (Compounds 26–39) (Table 4). These SAR studiesshowed tolerance for rather bulky substituents in this posi-tion, as exemplified by the 3-phenylamino derivatives 26–


Compound REGFR

(IC50 mM)

EGFELISA1

(IC50 mM)MK Cell2(IC50 mM)

15 p-OCH3 0.013 0.002 0.2916 p-NH2 0.002 0.010 0.2617 p-NHCOCH3 0.004 0.001 0.3418 p-NHSO2CH3 0.005 0.006 0.1919 p-CONH2 0.005 0.067 0.1820 m-NH2 0.001 0.001 0.1921 m-NHSO2CH3 0.004 0.010 0.8922 p-CONH2 0.005 0.067 0.18

1Inhibition of EGF-stimulated phosphorylation in A 431 cells.2Inhibition of proliferation of EGF-dependent BALB/MK cells.

FIGURE 4. Rational design of pyrazolo-pyrimidines.


30, the benzylamino derivatives 31–34, or Compounds35–39, where a phenyl ring is directly attached to the pyra-zole ring. The most potent compounds of this series hadIC50 values below 10 nM against the EGFR enzyme and aselectivity ratio .1000 against most tested tyrosine andserine/threonine kinases. In cells, EGF-stimulated cellulartyrosine phosphorylation was inhibited at IC50 values below50 nM, whereas PDGF-induced tyrosine phosphorylationwas not affected (IC50 .10 mM). In EGF-dependent MKcells, proliferation was blocked with IC50 values between0.4 and 1 mM).

Interestingly, the p-NHBoc derivatives 30 (IC50 .10mM) and 37 (IC50 .0.270 mM) had markedly decreasedactivity, indicating that the pocket is limited in size. Dock-ing analyses of Compound 27 in the ATP-binding site ofthe homology-built model of the EGFR kinase suggests ahydrogen bond interaction between the hydroxy group ofthe phenol moiety and the backbone amide group of resi-due Phe832. This amino acid corresponds to Phe185 in theX-ray structure of the cAMP-dependent protein kinase(Zheng et al., 1993), and is known to form a hydrogen bondwith a water molecule. It is suggested that expulsion and re-placement of this buried water molecule by a hydroxy oramino substituent of the inhibitor could provide additionalbinding affinity (Traxler et al., 1998).

Further biological profiling of selected 4-phenylamino-pyrazolo[3,4-d]pyrimidines is currently ongoing at Novartis.

6. ISOFLAVONES AND3-PHENYL-4(1H)-QUINOLONES

A quite recent successful application of our pharmacophoremodel led to a series of isoflavones and 3-phenyl-4(1H)qui-nolones as potent and selective EGFR tyrosine kinase in-hibitors (Traxler et al., 1999). Several flavonoid com-

pounds have been reported to possess protein kinaseinhibitory activity. The most prominent of these are theisoflavone genistein, the flavone quercetin, and flavopiridol(Fig. 5). Due to its substantial potency (micromolar) in in-hibiting the EGFR PTK, we considered genistein as a possi-ble lead structure and, therefore, were interested in model-ing its kinase-binding mode. The three compounds share a5-hydroxy-chromenone substructure, suggesting similarbinding modes. However, when we analyzed the publishedX-ray data of quercetin in complex with the Hck tyrosinekinase (Sicheri et al., 1997) and of deschloro flavopiridol incomplex with CDK2 (Filgueira de Azevedo et al., 1996), itwas not possible to dock genistein in the ATP-binding siteof the EGFR kinase in analogous ways by giving its 5-hy-droxy-chromenone moiety the same orientations as thoseseen in the above complexes. In both cases, we observed se-vere steric clashes between the p-hydroxyphenyl moiety inthe 3-position and protein residues of the hinge region, sug-gesting the existence of a different binding mode specific togenistein. A clue to the possible nature of this alternativebinding mode came by comparing genistein to the pheny-lamino-quinazoline inhibitor class reported by Fry et al.(1994) and Rewcastle et al. (1995). In the literature, wefound an example of enzyme inhibitors where a salicylicgroup has been reported to be a successful bioisosteric re-placement of a quinazoline moiety (Hodge and Pierce,1993). Thereby, the pyrimidine ring mimics the pseudo six-membered ring resulting from the hydroxy-keto intramo-lecular hydrogen bond in salicylic acid. A similar hydroxy-keto function is present in genistein. Assuming the samebioisosteric relationship between the quinazoline-typeEGFR kinase inhibitors and genistein led to the hypothesisthat they share a common binding mode. In Fig. 6, it can beseen that overlapping the 5-hydroxy-keto system ofgenistein and the pyrimidine ring of 4-(m-chlorophenyl-

TABLE 4. SAR of Pyrazolo-Pyrimidines

Compound REGFR

(IC50 mM) Compound REGFR

(IC50 mM) Compound REGFR

(IC50 mM)

26 m-Cl 0.033 31 H 0.007 35 p-OH 0.00627 m-OH 0.001 32 m-Cl 0.026 36 p-NH2 0.00228 m-OCH3 0.008 33 m-OCH3 0.008 37 p-NHBoc 0.27029 p-NH2 0.005 34 m-NH2 0.003 38 m-OH 0.02630 p-NHBoc .10 39 m-NH2 0.005


amino)-6,7-dimethoxy-quinazoline gives a very good over-all superimposition of the two molecules. In particular, theanilino moiety of the quinazoline superimposes nicely onthe p-hydroxyphenyl group of genistein. Given the impor-tance of the anilino meta-chloro (or m-bromo) substituentin conferring high EGFR PTK inhibitory activity to thequinazoline derivatives, as well as to the other classes ofphenylamino-pyrimidine inhibitors (e.g., phenylamino-pyr-rolo-pyrimidines and phenylamino-pyrazolo-pyrimidines),the overlay immediately suggested that replacing the p-hy-droxyphenyl moiety of genistein by a m-chlorophenyl ringwould enhance its potency. This idea motivated the syn-thesis of Compound 40 (Table 5) as a target structure,which in fact, with an IC50 value of 95 nM, was 10 timesmore potent than genistein (IC50 1 mM).

The proposed putative binding mode of 40 in the ATP-binding site of the EGFR kinase model is fully consistentwith our pharmacophore model (Fig. 6). The chlorophenylring fits in the “sugar pocket,” making a sulfur-aromatic in-teraction with Cys773. Mimicking ATP, the 5-hydroxysubstituent accepts a hydrogen bond from the backbone

NH of residue Met769, while the 7-hydroxy substituentdonates a hydrogen bond to the backbone carbonyl ofGln767. In accordance with the predicted binding mode,removal of the 5-hydroxy group in Compound 43 or itsmethylation (together with the 7-hydroxy group) in Com-pound 42 led to a more than 500-fold loss of enzyme inhib-itory activity, thus proving the absolute requirement of thisgroup (Table 5).

It should be noted that 40 can also be docked in theATP-binding site of the EGFR kinase according to theParke-Davis model (i.e., with the chlorophenyl ring fittinghydrophobic region I of our general kinase model, the5-hydroxy group still accepting a hydrogen bond from thebackbone NH of Met769). However, in this orientation,one cannot find a hydrogen bond acceptor on the proteinto be a partner for the 7-hydroxy group whose hydrogenbond donor functionality clearly contributes to affinity, astestified by the reduced activity of the methoxy analogue 41.

As predicted, replacement of oxygen O1 by a nitrogen,which, according to the model, is facing the part of the cav-ity binding the triphosphate chain of ATP, further im-

FIGURE 5. Structures of genistein, quercetin, flavopiridol, and quinazoline.

FIGURE 6. Superimposition of genistein (dark lines) with quinazoline (light lines).


proved potency by a factor of more than 10. With IC50 val-ues of 38 and 8 nM, respectively, the quinolones 44 and 45were the most potent compounds of the series (Table 5).Again, and in accordance with the proposed binding mode,removal of the 5-hydroxy group (Compound 46) com-pletely abolished enzyme inhibitory activity.

The activity profiles of the isoflavone 40 and especiallyof the quinolones 44 and 45 show that our pharmacophoremodel of the ATP-binding site, together with additional ra-tional ideas, can be used successfully for the design of novelclasses of highly potent and selective EGFR tyrosine kinaseinhibitors.

7. PHENYLAMINO-PYRIMIDINES

Although the following example of an ATP competitiveinhibitor was not identified with the help of the pharma-cophore model, it adds to the fact that the ATP-bindingsite of protein kinases is an exciting target and that minormodifications of the structure of a molecule can lead to ma-

jor changes in the selectivity profile. In the course of a ran-dom-screening program, Compound 47 was identified as anattractive lead structure, acting as an unselective inhibitoragainst the serine/threonine kinase PKC a in the submicro-molar range. Simply through introduction of a methylgroup in the 6-position of the phenyl ring, the inhibitionagainst PKC a was completely lost, whereas the potency forinhibition of PDGFR autophosphorylation was dramati-cally increased (Fig. 7). The dramatic loss of activityagainst PKC a was explained by a forced change of the pre-ferred conformation upon the introduction of this “flag-methyl” group, leading to severe clashes with the wall ofthe ATP-binding pocket within the PKC family of en-zymes. The optimization program of the lead structure 47finally led to Compound 48 (CGP 53716) and its water-soluble analogue 49 (CGP 57148). Both compounds weredual inhibitors of the v-Abl/PDGFR tyrosine kinases. CGP53716 inhibited PDGFR autophosphorylation and thev-Abl tyrosine kinase with IC50s of 100 and 400 nM, re-spectively, whereas CGP 57148 was more potent against

TABLE 5. SAR of Isoflavones and Quinolones

Compound R1 R2

EGFR(IC50 mM) Compound R1 R2

EGFR(IC50 mM)

40 OH OH 0.095 44 OH OCH3 0.03841 OH OCH3 1.27 45 OH OH 0.00842 OCH3 OCH3 500 46 H H .10043 H OH 58.8

FIGURE 7. Phenylamino-pyrimidines.


the v-Abl kinase (IC50 33 nM) than the PDGFR kinase(IC50 300 nM). Both compounds showed 100- to 1000-foldselectivity against a panel of over 20 serine/threonine andtyrosine kinases (including EGFR, IGF-IR, IR, and c-Src).Based on its more favourable physico-chemical profile,CGP 57148 (49) was selected for further development. Incells, v-Abl and PDGFR autophosphorylation, as well asPDGF-induced c-fos mRNA expression, were blocked withIC50s between 100 and 300 nM. Antiproliferative activityof CGP 57148 against v-Abl- or v-Sis-transformed cell lines,but not against EGF-dependent MK or FDC-P1 cells (IL-3-dependent), and a 60–80% decrease in the number of Bcr-Abl colonies in a colony-forming assay of peripheral bloodor bone marrow from patients with chronic myelogenousleukemia, was observed. In vivo, CGP 57148 inhibited tu-mor growth in nude mice inoculated with tumorigenic Bcr-Abl-expressing cells. The compound recently has enteredPhase I studies in patients with chronic myelogenous leuke-mia (Beran et al., 1998; Buchdunger et al., 1996; Carroll etal., 1997; Deininger et al., 1997; Druker et al., 1996; Gamba-corti-Passerini et al., 1997; Zimmermann et al., 1996, 1997).

8. OUTLOOK

In the last few years, it has been demonstrated with numer-ous examples that the ATP-binding sites of protein kinasesare exciting targets for medicinal chemists. We have dem-onstrated with a number of examples that a hypotheticalpharmacophore model for binding of inhibitors at theATP-binding site of the EGFR tyrosine kinase can be usedsuccessfully for the design of highly potent and selectiveEGFR tyrosine kinase inhibitors. Independently, othergroups have also developed models for the binding of theirlead structures in the ATP pocket of tyrosine kinases, suchas quinazolines in the EGFR kinase (Palmer et al., 1997),pyrido[2,3-d]pyrimidines in the c-Src tyrosine kinase(Trumpp-Kallmeyer et al., 1998), or oxindoles in FGFR orVEGFR tyrosine kinases (Mohammadi et al., 1997), andused these models for the design of more potent inhibitors.X-ray data of more than 30 published structures of proteinkinases complexed with ATP or an ATP competitive in-hibitor enabled the refinement of these models and theability to adapt it to almost any new kinase chosen as a tar-get. Medicinal chemists now have tools in hand that willallow rapid three-dimensional screening of large compoundlibraries for the identification of hits and new lead struc-tures. Combinatorial chemistry and parallel synthesis prob-ably will accelerate the optimization process of such leadstructures. Nevertheless, to overcome the hurdles of in vivoefficacy in animal models, pharmacodynamic parameters(blood and tissues levels or metabolism), or toxicology, itstill needs serendipity and creativity of medicinal chemistsin the design of single compounds.

ReferencesBeran, M., Cao, X., Estrov, Z., Jeha, S., Jin, G., O’Brien, S., Tal-

paz, M., Arlinghaus, R. B., Lydon, N. B. and Kantarjian, H.

(1998) Selective inhibition of cell proliferation and BCR-ABLphosphorylation in acute lymphoblastic leukemia cells express-ing Mr l90,000 BCR-ABL protein by a tyrosine kinase inhibitor(CGP 57148). Clin. Cancer Res. 4: 1661–1672.

Bridges, A. J. (1995) The current status of tyrosine kinase inhibi-tors: do the diarylamine inhibitors of the EGF receptor repre-sent a new beginning? Exp. Opin. Ther. Patents 5: 1245–1257.

Buchdunger, E., Zimmermann, J., Mett, H., Meyer, T., Müller, M.,Druker, B. J. and Lydon, N. B. (1996) Inhibition of the Ablprotein-tyrosine kinase in vitro and in vivo by a 2-phenylami-nopyrimidine derivative. Cancer Res. 56: 100–104.

Burke, T. R. (1992) Protein-tyrosine kinase inhibitors. DrugsFuture 17: 119–131.

Carroll, M., Ohno-Jones, S., Tamura, S., Buchdunger, E., Zimmer-mann, J., Lydon, N., Gilliland, D. G. and Druker, B. J. (1997)CGP 57148B, a tyrosine kinase inhibitor, inhibits the growth ofcells expressing BCR-ABL, TEL-ABL, and TEL-PDGFR fusionproteins. Blood 90: 4947–4952.

Deininger, M. W., Goldman, J. M., Lydon, N. and Melo, J. V. (1997)The tyrosine kinase inhibitor CGP 57148B selectively inhibitsthe growth of Bcr-Abl-positive cells. Blood 90: 3691–3698.

Druker, B. J., Tamura, S., Buchdunger, E., Ohno, S., Segal, G. M.,Fanning, S., Zimmermann, J. and Lydon, N. (1996) Effects of aselective inhibitor of the Abl tyrosine kinase on the growth ofBcr-Abl positive cells. Nature Med. 2: 561–566.

Filgueira de Azevedo, W., Jr., Mueller-Dieckmann, H. J., Schulze-Gahmen, U., Worland, P. J., Sausville, E. and Kim, S.-H.(1996) Structural basis for specificity and potency of a flavanoidinhibitor of human CDK2, a cell cycle kinase. Proc. Natl.Acad. Sci. USA 93: 2735–2740.

Fry, D. W. (1994) Protein tyrosine kinases as therapeutic targets incancer chemotherapy and recent advances in the developmentof new inhibitors. Exp. Opin. Invest. Drugs 3: 577–595.

Fry, D. W., Kraker, A. J., McMichael, A., Ambroso, L. A., Nelson,J. M., Leopold, W. R., Connors, R. W. and Bridges, A. J. (1994)A specific inhibitor of the epidermal growth factor receptortyrosine kinase. Science 265: 1093–1095.

Furet, P., Caravatti, G., Lydon, N., Priestle, J., Sowadski, J., Trinks,U. and Traxler, P. (1995) Modeling study of protein kinaseinhibitors: binding mode of staurosporine—origin of the selec-tivity of CGP 52411. J. Comput. Aided Mol. Des. 9: 465–471.

Gambacorti-Passerini, C., Le Coutre, P., Mologni, L., Fanelli, M.,Bertazzoli, C., Marchesi, E., Di Nicola, M., Biondi, A., Corneo,G. M., Belotti, D., Pogliani, E. and Lydon, N. B. (1997) Inhibi-tion of the Abl kinase activity blocks the proliferation of Bcr/Abl1 leukemic cells and induces apoptosis. Blood Cells Mol.Dis. 23: 380–394.

Hanke, J. H., Gardner, J. P., Dow, R. L., Changelian, P. S., Bris-sette W. H., Weringer, E. J., Pollok, B. A. and Connelly, P. A.(1996) Discovery of a novel, potent, and Src family-selectivetyrosine kinase inhibitor. J. Biol. Chem. 271: 695–701.

Hodge, C. N. and Pierce, J. A. (1993) A diazine heterocyclereplaces a six-membered hydrogen-bonded array in the activesite of scytalone dehydratase. Bioorg. Med. Chem. Lett. 3:1605–1608.

Lawie, A. M., Noble, M. E. M., Tunnah, P., Brown, N. R.,Johnson, L. N. and Endicott, J. A. (1997) Protein kinase inhibi-tion by staurosporine revealed in details of the molecular inter-action with CDK2. Nature Struct. Biol. 4: 796–801.

Levitzky, A. and Gazit, A. (1995) Tyrosine kinase inhibition: anapproach to drug development. Science 267: 1782–1788.

Mohammadi, M., McMahon, G., Sun, L., Tang, C., Hirth, P.,


Yeh, B. K., Hubbard, S. R. and Schlessinger, J. (1997) Struc-tures of the tyrosine kinase domain of fibroblast growth factorreceptor in complex with inhibitors. Science 276: 955–959.

Palmer, B. D., Trumpp-Kallmeyer, S., Fry, D. W., Nelson, J. M.,Showalter, H. D. H. and Denny, W. A. (1997) Tyrosine kinaseinhibitors. 11. Soluble analogues of pyrrolo- and pyrazolo-quinazolines as epidermal growth factor receptor inhibitors:synthesis, biological evaluation, and modeling of the mode ofbinding. J. Med. Chem. 40: 1519–1529.

Rewcastle, G. W., Denny, W. A., Bridges, A. J., Zhou, H., Cody,D. R., McMichael, A. and Fry, D. W. (1995) Tyrosine kinaseinhibitors. 5. Synthesis and structure-activity relationships for4-(phenylmethyl-amino)- and 4-(phenylamino)quinazolinesas potent adenosine 59-triphosphate binding site inhibitors ofthe tyrosine kinase domain of the epidermal growth factorreceptor. J. Med. Chem. 38: 3482–3487.

Rewcastle, G. W., Palmer, B. D., Thompson, A. M., Bridges, A. J.,Cody, D. R., Zhou, H., Fry, D. W., McMichael, A., Kraker,A. J. and Denny, W. A. (1996) Tyrosine kinase inhibitors. 10.Isomeric 4-[(3-bromophenyl)-amino]-pyrido[d]-pyrimidines arepotent ATP binding site inhibitors of the tyrosine kinase func-tion of the epidermal growth factor receptor. J. Med. Chem. 39:1823–1835.

Rewcastle, G. W., Bridges, A. J., Fry, D. W., Rubin, J. A. andDenny, W. A. (1997) Tyrosine kinase inhibitors. 12. Synthesisand structure-activity relationships for 6-substituted-4-(phenyl-amino)pyrimido[5,4-d]pyrimidines designed as inhibitors of theepidermal growth factor receptor. J. Med. Chem. 40: 1820–1826.

Schulze-Gahmen, U., Brandsen, J., Jones, H. D., Morgan, D. O.,Meijer, L., Vesely, J. and Kim, S. H. (1995) Multiple modes ofligand recognition: crystal structures of cyclin-dependent pro-tein kinase 2 in complex with ATP and two inhibitors, olo-moucine and isopentenyladenine. Proteins 22: 378–391.

Sicheri, F., Moarefi, I. and Kuriyan, J. (1997) Crystal structure ofthe Src family tyrosine kinase Hck. Nature 385: 602–609.

Spada, A. P. and Myers, M. R. (1995) Small molecule inhibitorsof tyrosine kinase activity. Exp. Opin. Ther. Patents 5: 805–817.

Thompson, A. M., Bridges, A. J., Fry, D. W., Kraker, A. J. andDenny, W. A. (1995) Tyrosine kinase inhibitors. 7. 7-Amino-4-(phenylamino)- and 7-amino-4-(phenyl-methyl)amino-pyrido[4,3-d]pyrimidines: a new class of inhibitors of thetyrosine kinase activity of the epidermal growth factor receptor.J. Med. Chem. 38: 3780–3788.

Traxler, P. (1997) Protein tyrosine kinase inhibitors in cancertreatment. Exp. Opin. Ther. Patents 7: 571–588.

Traxler, P. (1998) Protein tyrosine kinase inhibitors in cancertreatment (part II). Exp. Opin. Ther. Patents.

Traxler, P., Green, J., Mett, H., Séquin, U. and Furet, P. (1999)Use of a pharmacophore model for the design of EGFR tyrosinekinase inhibitors: isoflavones and 3-phenyl-4(1H)-quinolones.J. Med. Chem., in press.

Traxler, P. and Lydon, N. (1995) Recent advances in proteintyrosine kinase inhibitors. Drugs Future 20: 1261–1274.

Traxler, P., Furet, P., Mett, H., Buchdunger, E., Meyer, T. andLydon, N. (1996) 4-(Phenylamino)pyrrolopyrimidines: potentand selective, ATP site directed inhibitors of the EGF-receptorprotein tyrosine kinase. J. Med. Chem. 38: 2285–2292.

Traxler, P., Bold, G., Frei, J., Lang, M., Lydon, N., Mett, H.,Buchdunger, E., Meyer, T., Mueller, M. and Furet, P. (1997)Use of a pharmacophore model for the design of EGF-Rtyrosine kinase inhibitors: 4-(phenylamino)pyrazolo-pyrim-idines. J. Med. Chem. 40: 3601–3616.

Traxler, P., Bold, G., Buchdunger, E., Lang, M., Mett, H. andFuret, P. (1998) The use of a pharmacophore model for thedesign of potent and selective EGFR tyrosine kinase inhibitors.In: 215th American Chemical Society National Meeting, Dal-las, TX, 29 March-2 April, Lecture, Medi 104.

Trumpp-Kallmeyer, S., Rubin, J. R., Humblet, C., Hamby, J. M.and Showalter, H. D. H. (1998) Development of a bindingmodel to protein tyrosine kinases for substituted pyrido[2,3-d]pyrimidine inhibitors. J. Med. Chem. 41: 1752–1763.

Ward, W. H. J., Cook, P. N., Slater, A. M., Davies, D. H., Hold-gate, G. A. and Green, L. R. (1994) Epidermal growth factorreceptor tyrosine kinase. Investigation of catalytic mechanism,structure-based searching and discovery of a potent inhibitor.Biochem. Pharmacol. 48: 659–666.

Zheng, J., Knighton, D. R., Ten Eyck, L. F., Karlsson, R., Xuong,N. H., Taylor, S. S. and Sowadski, J. M. (1993) Crystal struc-ture of the catalytic subunit of cAMP-dependent protein kinasevomplexed with MgATP and peptide inhibitor. Biochemistry32: 2154–2161.

Zimmermann, J., Buchdunger, E., Mett, H., Meyer, T., Lydon, N.and Traxler, P. (1996) Phenylamino-pyrimidine (PAP)-deriva-tives: a new class of potent and highly selective PDGF-receptorautophosphorylation inhibitors. Bioorg. Med. Chem. Lett. 6:1221–1226.

Zimmermann, J., Buchdunger, E., Mett, H., Meyer, T. and Lydon,N. (1997) Potent and selective inhibitors of the Abl-kinase:phenylaminopyrimidine (PAP)-derivatives. Biorg. Med. Chem.Lett. 7: 187–192.

associate editor: d. shugar strategies toward the design ... · model led to the identification and...

Documents